Time to add the epicycles!

During the golden age of macroeconomics (roughly 1984-2007), many economists understood that interest rates were not monetary policy. After 2008, economists have been drifting back to old-school Keynesianism, with its emphasis on fiscal policy and interest rates.

For the umpteenth time, it makes no sense to talk about interest rates causing changes in other macro variables. To do so is to engage in reasoning from a price change. Interest rates can change for multiple reasons, and the effects of the rate change will depend on the underlying factors that caused rates to change.

Not surprisingly, this highly flawed approach has produced lousy results. The Financial Times reports that dissatisfaction with our current models has led to the search for alternatives:

Matthew Rognlie of Northwestern University says that more broadly, the Hank trend tapped into a “well of discontent” with older, simpler models. Those assume consumers respond very strongly to changes in interest rates, and hardly at all to changes in their income.

I cannot say I’m surprised by the fact that models that assume consumers respond to interest rate changes in a predictable way have not done well. Unfortunately, instead of scrapping the interest rate approach to macroeconomics (which I recommend in my recent book) economists are adding epicycles:

Hank [Heterogeneous Agent New Keynesian] models try to match real-life spending behaviour more closely, assuming a willingness to consume out of extra income roughly 10 times larger than in the older models.

That changes the emphasis when thinking about monetary policy transmission, away from the idea that greater rewards for saving encourage more of it. Other mechanisms could include an interest rate increase that hits people with variable-rate mortgages living hand-to-mouth, damping spending. Or an interest rate cut could stimulate investment, pumping up wages of people who are particularly likely to splurge, boosting consumption.

Another way that economists handle the failures of modern macro is by making the predictions more ambiguous, a technique used by successful astrologers. For instance, they can invoke those mysterious “long and variable lags”:

Despite including more detail, there are still areas where such models don’t seem to meet a reality check. They don’t capture the fact that individual spending can take a while to respond to an interest rate change.

High interest rates don’t reduce aggregate demand? You just wait. It must be those long and variable lags. The phrase “a while” is so much better than “6 months” or “18 months” or “30 months”. In early 2023, economists told us the recession was delayed because of long and variable lags. OK, but for how long? It’s already April 2024; is the recession coming soon?

Perhaps the following analogy would be useful: How do rising oil prices affect consumption, other things equal? That’s not even a question. Other things equal, oil prices never change. If oil prices rise due to reduced supply, then consumption falls. If oil prices rise because of increased demand, then consumption rises. But other things equal? What does that even mean?

If interest rates rise because of tight money, then aggregate demand may decline. If interest rates rise because of fiscal deficits or booming immigration or strong “animal spirits”, then aggregate demand may rise. It depends.

Doesn’t the Fed determine interest rates? Well, it has a target, which it moves up and down in response to what it perceives as changes in the equilibrium interest rate. But is it leading the market, or following?

No doubt the defenders of these models will insist that they’ve already incorporated all the various factors that move the equilibrium rate of interest. All I can say is that the proof is in the pudding—apparently we are still not able to model that “natural” rate with any degree of accuracy. As a result, we end up reasoning from a price change.

Monetary policy is not interest rates, it is the market forecast of future NGDP.



23 Responses to “Time to add the epicycles!”

  1. Gravatar of copans copans
    12. April 2024 at 11:34

    I’m going to go out on a limb here, but I think it is folly to reason from a price change.

  2. Gravatar of Solon of the East Solon of the East
    12. April 2024 at 15:55

    If you are old enough to remember fortran cards, you remember when even right wingers said that any rate of inflation under five percent is good enough.

    Perhaps the current macroeconomic policies are ok. I am not reasoning from the level of interest rates for all the reasons Sumner has illuminated.

    I do wonder about quantitative tightening. By selling trillions in US Treasuries, the Federal Reserve will suck a minute percentage of money out of global capital markets.

    Will that affect inflation rates which are moderate anyway?

    But US taxpayers will be leveraged up going forward.

    Are there lessons from the Bank of Japan in this situation?

  3. Gravatar of Rajat Rajat
    12. April 2024 at 16:46

    In Australia, many people claim that higher official interest rates affect spending more directly than in the US due to our much larger proportion of home loans with variable rates. See here for instance: https://x.com/Scutty/status/1752549029685481761 This they say is the reason why the RBA should cut official rates before the Fed. Of course, the same could have be said in 2007, but the RBA nevertheless (rightly) moved to cut later than the Fed because our nominal spending was much stronger (NGDP growth was then around 10% pa). People keep committing Friedman’s interest rate fallacy! of course, none of this is to say that the RBA shouldn’t cut before the Fed – I think it should and will – but the reason is not the structure of our mortgages.

    Incidentally, would you also say something similar about Gauti Eggertsson and Pierpaulo Benigno’s non-linear New Keynesian Phillips Curve (see https://www.nber.org/papers/w31197) – that it is a reaction to “a well of discontent” with the PC as a causal relationship between the labour market and inflation?

  4. Gravatar of Solon of the East Solon of the East
    13. April 2024 at 00:27

    OT on unhappy Americans despite better economy: A map shows strong GDP growth in a lot of non-coast states such as Texas.

    Some people, in Texas, for example, are facing much higher housing costs (this applies across the entire West Coast).

    Along the coasts, the economy is not growing as fast, and cost of living (housing) is eviscerating household budgets.

    This leaves a small number of non-coast states where housing is affordable and GDP is going up.

  5. Gravatar of ssumner ssumner
    13. April 2024 at 04:24

    Rajat, Good point about dissatisfaction with the PC.

    On interest rates, I’d add that in America mortgages tend to be linked to long-term rates, for which the “neofisherian” argument is especially strong. The UK is different—don’t know about Australia.

  6. Gravatar of spencer spencer
    13. April 2024 at 05:47

    The movement of funds through the nonbanks increases the supply of loan funds, but not the supply of money and vice versa.

    The NBFIs are the customers of the DFIs. I.e., lending by the NBFIs is noninflationary, other things equal. It is a velocity relationship activating monetary savings.

  7. Gravatar of Todd Ramsey Todd Ramsey
    13. April 2024 at 05:51

    Did you intend to write “scrapping”, not “scraping”?

    Do you think the focus on interest rates as monetary policy is because people are still using the gold standard model of the central bank adjusting interest rates to adjust the amount of gold on deposit? And just not being cognizant that fiat necessitates a whole new way of thinking? Or if not, why do you think people remain focused on interest rates?

  8. Gravatar of spencer spencer
    13. April 2024 at 05:56

    The NBFIs are not in competition with the DFIs. And unless monetary savings are activated a dampening economic impact is generated (secular stagnation).

  9. Gravatar of ssumner ssumner
    13. April 2024 at 06:02

    Todd, That might be part of it. I’d add that there wasn’t much of a Fisher effect under the gold standard–it was mostly the liquidity effect (and the income effect.)

    “Did you intend to write “scrapping”, not “scraping”?”

    That too. 🙂

  10. Gravatar of spencer spencer
    13. April 2024 at 06:58

    N=gDp has increased by 40 percent since 2qt/20

  11. Gravatar of Cameron Blank Cameron Blank
    13. April 2024 at 07:38

    Funny, I accidently clicked something on your blog and ended up on a 2012 post of you responding to Krugman on the radio. Back then it seemed almost impossible to get more than 200k jobs a month and push inflation up to 2% whereas now it seems impossible to get less than 300k jobs/month and drag inflation down below 3%.

    Do you have any idea what may have caused this drastic difference? What kind of monetary regime may have smoothed this out?

  12. Gravatar of Cameron Blank Cameron Blank
    13. April 2024 at 07:45

    Kidding of course.

    We all know the fed has limited ammo to fight inflation now just as it couldn’t do anything to push it up in 2008-2014. If only the fed could raise the FF rate above 5.5%…

  13. Gravatar of Scott Sumner Scott Sumner
    13. April 2024 at 07:47

    Cameron, The funny part is that I thought we were almost there with FAIT, until I found out the Fed intended it to be asymmetrical

  14. Gravatar of Bobster Bobster
    13. April 2024 at 08:17

    It’s unfortunate that our monetary policy seems to be going backwards after apparently learning mistakes from the Great Recession.

    Oddly, it was average inflation targeting that may have done them in, along with the delay in the Fed chair pick which then delayed the hiking cycle.

    There’s also a lot of motivated reasoning to defend Biden in an election year.

  15. Gravatar of Eharding Eharding
    13. April 2024 at 09:57

    “It’s already April 2024; is the recession coming soon?”

    I think so, Sumner; I expect a recession this year (or at least by February 2025).

  16. Gravatar of Iskander Iskander
    14. April 2024 at 05:14

    It is strange to observe how they are making the New Keynesian model, which very few people seem to properly understand (having read John Cochrane and Nick Rowe, I don’t think I really do), even more complicated. May as well add another couple of years to graduate programs – even though you still won’t learn that interest rates aren’t monetary policy.

  17. Gravatar of lysseas lysseas
    14. April 2024 at 06:50

    “Unfortunately, instead of scrapping the interest rate approach to macroeconomics (which I recommend in my recent book) economists are adding epicycles”: Like listening to Sabine talking about String Theory physicists.

  18. Gravatar of Michael Sandifer Michael Sandifer
    14. April 2024 at 19:53

    I’m guessing the Fed will move to a symmetric average inflation target next year.

  19. Gravatar of msgkings msgkings
    15. April 2024 at 14:18

    @Michael S:

    What makes you guess that? Not challenging you, just wondering. It doesn’t seem like something they are comfortable with or they would have already…

  20. Gravatar of ssumner ssumner
    15. April 2024 at 15:38

    Michael, I doubt that, but I hope you are right.

  21. Gravatar of Philo Philo
    15. April 2024 at 21:25

    ”But other things equal? What does that even mean?” Good question! If there is no clear answer, most–even all–counterfactual statements and questions are doubtfully meaningful. They have the form: “If Not-P were true [P being the actual truth], Q would be true [or “. . . would Q be true?”], *ceteris paribus*.” If we can’t cash out that “*ceteris paribus*, we cannot understand the claim/question.

  22. Gravatar of William Peden William Peden
    16. April 2024 at 02:36

    Oddly enough, I was in a philosophy reading group talking about this type of question, but in a medical context. Consider: does obesity cause death? Some epidemiologists argue that such a question is ill-specified. Different interventions on obesity (e.g. diet vs. exercise vs. both) will have different effects on mortality risk. Different ways of becoming obese (e.g. genetics vs. lifestyle vs. both) will have different effects too.

    You can ask, “What if all things were equal and someone became more obese?” but obesity causally affects other things (including diet) so this question is also not well-specified.

    These epidemiologists argue that, instead, the scientifically answerable question is we can ask, “For this well-defined intervention (e.g. reducing obesity through this particular diet or raising 2-year NGDP expectations) what is the difference made on this well-defined outcome?”

  23. Gravatar of ssumner ssumner
    16. April 2024 at 06:41

    Philo, The economic claims of interest are policy counterfactuals.

    William, Good analogy.

Leave a Reply